Ok, I think you might have missed the point of my original email, so I'll try to re-state it :
1) If people have lots of choices, they would rather not make a choice, particularly when they don't have the knowledge to make a value judgment as to what the "best" choice is.
In this case, there is no value judgment to be made, but with say 100 - 1000 (or many more) EUI-48 addresses to choose from (remember, each switch has a EUI-48 address per port), it will *feel* like there is a value judgement to be made. And that will confuse them.
Too many choices can almost be as bad as not enough.
I'm pretty sure I get your point, I just don't buy it. This isn't a complex decision for the user and software can help greatly.
It doesn't matter which one, software can pick one from the box you are performing the configuration on.2) Even if you automate the process of generating a local range address using a EUI-48 MAC address, which device do you perform it on ? For example, with 200 routers, and multiple Internet connections (ie. so you can't say the "Internet" router is the one to get an EUI-48 from), again, you can have too many choices. Which router / interface is the "right" EUI-48 source ?
3) Alternatively, if there was a manual EUI-48 entry prompt, to allowWhy would there be an EUI-48 entry prompt? There is a need to configure/propagate around the hashed prefix that resulted from the computation, exactly as in the birthdate approach.
people to select an arbitrary EUI-48 from their set, if they don't know
how to find out an EUI-48 address, and / or don't understand the value
in selecting a IEEE assigned EUI-48, they will enter 00:00:00:00:00:00,
or 11:11:11:11:11:11, etc.
These are people / technology issues, not just technology issues.I don't think there are any "problems" with using an EUI-48 as an input to the hash that are avoided by the birthdate. And it has the advantages of being able to be performed automatically, and of providing "good uniqueness" of bits.
Bob's use of a ntp time stamp and the operator's birthday as the hash
input avoids these problems.
Another approach might be to hash the admin's phone number. (Ideally we could staple a whole phone number into the prefix, which would be cute from a debugging point of view. Unfortunately with 41 bits we only get 12 digits.)
- aidan
-------------------------------------------------------------------- IETF IPng Working Group Mailing List IPng Home Page: http://playground.sun.com/ipng FTP archive: ftp://playground.sun.com/pub/ipng Direct all administrative requests to [EMAIL PROTECTED] --------------------------------------------------------------------
